TARQUIN or PRESS parameter optimization???

[harshal patel]

Dear TARQUIN experts,

I do single voxel spectroscopy to investigate GABA concentration using a method described by Napolitano et al(TE 105ms).  We have a new Siemens PRISMA scanner and I use svs_se protocol provided by the Siemens to acquire the spectra. After running the tarquin analyses with default parameters I donot see reliable gaba concentration. Is there something wrong with the TARQUIN parameters or do I need to change Siemens protocol or the voxel positioning??
I acquired proton spectra from four participants untill now.

In each measurement I acquire water supperessed and unsuppressed spectra separately 
The contrast parameters for water suppressed and water unsupressed spectra are as follow.

Water suppressed

Water suppressed = yes
Water suppression bandwidth = 50 hz
Spectral suppression = fat and water suppression
Fat suppression bandwidth = 1.55 ppm
Fat suppression delta position = -3.40 ppm
Water suppression bandwidth = 1.55ppm
Water suppression = 0.00 ppm

Water unsuppressed
Water suppression = No
Spectral suppression = No

Voxel Size 3x3x3 cm

The TARQUIN analyses and the voxel placement is shown in the attached images.
V1,V2, V3 and V4 are the voxel positions from First, second, third and fourth participants and S1, S2 ,S3 and S4 are the spectra from the corresponding participants respectively.

Any help will be highly appreciated.


Many thanks for your precious time and efforts

Best,
Harshal

[Martin Wilson]

Dear Harshal,

You have some very odd spectral artefacts here. Are you exporting data
that has been preprocessed by the Siemens software?

Also you are very close to the edge of the brain (too close really).
You could try orientating the voxel so the uppermost edge is parallel
with the skull.

Martin

Martin

> --
> You received this message because you are subscribed to the Google Groups
> "TARQUIN users' group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to tarquin_users_g...@googlegroups.com.
> To post to this group, send email to tarquin_u...@googlegroups.com.
> Visit this group at http://groups.google.com/group/tarquin_users_group.
> For more options, visit https://groups.google.com/d/optout.

[Felix Raschke]

Hi Harshal,

I responded to your post this morning from my email account but somehow it didn't arrive at the google group. Anyway I repost my response below, Martin already mentioned the voxel placement and lipid artefacts.

A couple of points:

1. If you want to use the GABA PRESS 105 sequence suggested by Napolitano et al. you need to make sure your are using a PRESS sequence with TE=105 AND TE1=15ms. If you leave TE1 at its default value you are unlikely to identify GABA.
2. If you set up the sequence as in 1. then make sure you adjust the TE1 in the TARQUIN data parameters of the advanced fit to 15ms also!
3. Your voxel placement is not optimal. The voxels are positioned to close to the skull and due to the large chemcial shift artefact at 3T you will get large lipid contamination in your spectra if the chemcial shift for lipids is in the direction of the skull. A brief drawing attached to clarify. Please note that you should be able to adjust the chemical shift direction --> make sure lipid excitation volume is shifted away from the skull.

hope that helps,

 

best wishes

 

Felix

[harshal patel]

Dear Martin,
Dear Felix,

Thank you so much for taking time and giving your expert comments on the spectroscopy data.

@Martin: Yes I have used spectral suppression in the protocol and it might have preprocessed the data before exporting. I would switch it off next time.
@ Martin & Felix: Agree with the voxel placement.
@Felix: We have not set-up the sequence like you mentioned in your  first point. In fact siemens donot have any svs_se protocol with TE1 or TE2 options. Do we have to ask for a additional protocol from siemens or could we program it? I could do programming as I have been working with different programming languages. Soa little hint/support would be enough incase its a matter of progamming.
Third point: I didnot get when you said "make sure lipid excitation volume is shifted away from the skull"  Could you please elaborate?? How could we make lipid volume shift away from the skull??

Thank you Martin and Felix once again.

Best,
Harshal

[Felix Raschke]

Hi Harshal,

regarding TE1 adjustment on the Siemens, there must be a way of doing this. I don't have any experience with Siemens scanners but I am sure people here will know how to change TE1 - maybe they can help. You might need to contact somebody from Siemens though. Just ask for TE1 adjustment in your MRS sequence, that should hopefully not be a problem. Either way you want to know the TE1 so TARQUIN can simulate the basis set more accurately (? Martin ?). A TE1 other than the suggested 15ms in the Napolitano paper might not work for GABA detection. Also remember this optimisation is done at 3T, so is unlikely to work at 1.5T.

regarding the "lipid excitation volume";
The MRS box you are placing in the planning window is only at that location for a specific chemical shift (e.g. at 2ppm for Philips or 2.68ppm for GE). For any other chemical shifts such as lipids (1.3ppm - 0.9ppm) the MRS box is shifted in all three dimensions. If you are close to the skull with MRS planning, it is important to know in which direction the lipid MRS box is shifted (e.g. (a), (b), or (c) in the attached example) - this way lipid artifact can be avoided by going for option (b) in the example. Unfortunately, I can't tell you how to visualize and change the chemical shift direction on the Siemens scanner - again others might be able to help with this.

best wishes

Felix